The control arm in this trial was the same air purifier with the filter pulled out. Same box, same hum (58 decibels instead of 60), same breeze at the vent. Participants lived with it for a month and could not tell. The blinding index says the disguise held.

So for one month, about a hundred households near a Massachusetts interstate ran the most expensive fans in Somerville in their bedrooms and living rooms. Then they got the real one. The question was whether their blood pressure noticed.

HAFTRAP (the HEPA Filtration for Traffic Related Air Pollution trial) enrolled adults 30 and older living within 200 meters of I-93 or Route 38. Each household got one month of true HEPA filtration and one month of sham, in random order, with a one month washout between. Blood pressure was measured at home on weekday mornings at the start and end of each period.

The exclusion list is long and sensible for a mechanism study. No smokers or vapers, no prior heart attack or stroke, nobody on blood pressure or anti-inflammatory medication, nobody with a combustion source at home beyond the stove. The resulting group averaged 41 years old with a baseline pressure of 118.8/76.5. Most were college educated and earned more than $85,000 a year. These were healthy people who happen to live next to a highway.

The senior figure is Doug Brugge, now at UConn Health, formerly Tufts. He has studied near highway pollution in Somerville and Boston's Chinatown for more than fifteen years through the CAFEH project, a community partnership, and he wrote a book on the subject titled Particles in the Air: The Deadliest Pollutant Is One You Breathe Every Day. The subtitle tells you where he stands before the first page. Misha Eliasziw at Tufts is the statistician on both papers. Two coauthors on the JACC paper come from Somerville community groups (the Welcome Project and the Somerville Transportation Equity Partnership). The paper says plainly that the trial answered neighborhood requests to test protective measures.

Funding came from NIEHS. Austin Air custom built the units and sold them at a bulk discount, and the cognition paper states the manufacturer had no role in design or analysis. I see no financial conflict worth worrying about.

The bias here is intellectual and civic. A group that has spent a career documenting a harm, alongside the neighbors who live with it, wants a remedy to work. That does not make the data wrong. It does explain why the JACC discussion ends by recommending purifiers for vulnerable populations on the strength of a subgroup, and why the cognition paper leads with the one test and one age band that moved.

The American Heart Association laid out the pathway in a 2010 scientific statement, and it has held up. Inhaled particles irritate the airway and alveoli. Receptors there trigger an autonomic reflex that tips the balance toward sympathetic tone. Oxidative stress and inflammatory mediators spill into the circulation, nitric oxide becomes less available, and arterioles constrict. Blood pressure rises within hours of exposure, and controlled exposure studies show an effect in an afternoon.

The brain story is less settled. Ultrafine particles can travel along the olfactory nerve or cross a leaky blood brain barrier, and systemic cytokines can prime microglia. Same inflammatory machinery, different target organ. It is plausible, and nobody has shown the dose that matters.

So HAFTRAP is best read as a test of turning the dial down. If particles raise pressure through a known reflex, removing particles at home should lower it a little in people with some pressure to spare. That is roughly what happened.

On blood pressure, HAFTRAP lands where the field already was. A 2023 meta-analysis of 17 true versus sham trials (about 880 people) found a pooled systolic reduction of 2.35 mmHg, rising to 3.62 when high bias studies were dropped. The trials were small, and most of the big effects came from Chinese cities with far dirtier air than Somerville. A 3 mmHg shift is modest for one patient. Across a population it is the kind of number Geoffrey Rose built the prevention paradox on.

On infection the data are softer. A 2023 systematic review of 32 studies found no convincing evidence yet that air cleaning reduces respiratory infection in real world settings. The best recent trial, HEPA units in Australian aged care rooms, missed significance in its main analysis (odds ratio 0.57, confidence interval 0.32 to 1.04) and reached it only among completers.

And yet. During COVID, the observational signal and the physics pointed the same way. Georgia elementary schools that improved ventilation had 39% lower incidence, and those combining dilution with HEPA filtration had 48% lower, after adjusting for county rates. Observational, confounded, and consistent with everything we know about aerosols. More air exchanges were better than fewer. Nobody needed a sham window to believe it.

The honest synthesis is that clean air reliably reduces what you breathe, and the clinical endpoints downstream are small, noisy, and hard to measure in a month. That is a description of most preventive medicine.